Biology subjects

Cui, X.

Publications and source records attributed to Cui, X..

At least 91 records · Page 5Linked to original sources

bioRxiv2024

HisCl1 regulates gustatory habituation in sweet taste neurons and mediates sugar ingestion in Drosophila

Similar to other animals, the fly, Drosophila melanogaster, reduces its responsiveness to tastants with repeated exposure, a phenomenon called gustatory habituation. Previous studies have focused on the circuit basis of gustatory habituation in the fly chemosensory system 1,2. However, gustatory neurons reduce their firing rate during repeated stimulation 3, suggesting that cell-autonomous mechanisms also contribute to habituation. Here, we used deep learning-based pose estimation and optogenetic stimulation to demonstrate that continuous activation of sweet taste neurons causes gustatory habituation in flies. We conducted a transgenic RNAi screen to identify genes involved in this process and found that knocking down Histamine-gated chloride channel subunit 1 (HisCl1) in the sweet taste neurons significantly reduced gustatory habituation. Anatomical analysis showed that HisCl1 is expressed in the sweet taste neurons of various chemosensory organs. Using single sensilla electrophysiology, we showed that sweet taste neurons reduced their firing rate with prolonged exposure to sucrose. Knocking down HisCl1 in sweet taste neurons suppressed gustatory habituation by reducing the spike frequency adaptation observed in these neurons during high-concentration sucrose stimulation. Finally, we showed that flies lacking HisCl1 in sweet taste neurons increased their consumption of high-concentration sucrose solution at their first meal bout compared to control flies. Together, our results demonstrate that HisCl1 tunes spike frequency adaptation in sweet taste neurons and contributes to gustatory habituation and food intake regulation in flies. Since HisCl1 is highly conserved across many dipteran and hymenopteran species, our findings open a new direction in studying insect gustatory habituation.

neuroscience↗

bioRxiv2024

Artesunate, EDTA and colistin work synergistically against MCR-negative and -positive colistin-resistant Salmonella

Discovering new strategies to combat the multi-drug resistance bacteria constitutes a major medical challenge of our time. Previously, artesunate (AS) has been reported to exert antibacterial enhancement activity in combination with {beta}-lactam antibiotics, via inhibition of the efflux pump AcrB. However, combination of AS and colistin (COL) revealed weak synergistic effect against a limited number of strains, and few studies have further explored its possible mechanism of synergistic action. In this paper, we found that AS and EDTA could strikingly enhance the antibacterial effects of COL against mcr-1- and mcr-1+ Salmonella strains either in vitro or in vivo, when used in triple combination. The excellent bacteriostatic effect was primarily related to the increased cell membrane damage, accumulation of toxic compounds and inhibition of MCR-1. The potential binding sites of AS to MCR-1 (THR283, SER284, and TYR287) were critical for its inhibition of MCR-1 activity. Additionally, we also demonstrated that the CheA of chemosensory system and virulence-related protein SpvD were critical for the bacteriostatic synergistic effects of the triple combination. Selectively targeting CheA, SpvD or MCR using the natural compound AS could be further investigated as an attractive strategy for treatment of Salmonella infection. Collectively, our work opens new avenues towards the potentiation of COL and revealed an alternative drug combination strategy to overcome COL resistant bacterial infections.

microbiology↗

bioRxiv2024

collectNET: a web server for integrated inference of cell-cell communication network

SummaryCell-cell communication through ligand-receptor pairs forms the cornerstone for complex functionalities in multicellular organisms. Deciphering such intercellular signaling can contribute to un-raveling disease mechanisms and enables targeted therapy. Nonetheless, notable biases and inconsistencies are evident among the inferential outcomes generated by current methods for inferring cell-cell communication network. To fill this gap, we developed collectNET (http://health.tsing-hua.edu.cn/collectnet) as the first web server for efficiently inferring the cell-cell communication network, with efficient calculation, hierarchical browsing, comprehensive statistics, advanced searching, and intuitive visualization. collectNET provides a reliable online inference service with prior knowledge of three public ligand-receptor databases and systematic integration of three mainstream inference methods. Additionally, collectNET has assembled a human cell-cell communication atlas, including 126,785 significant communication pairs based on 343,023 single cells. We anticipate that collectNET will benefit researchers in gaining a more holistic understanding of cell development and differentiation mechanisms. Availability and ImplementationcollectNET is freely available at http://health.tsinghua.edu.cn/collectnet. Contactruijiang@tsinghua.edu.cn Supplementary informationSupplementary data are available at Bioinformatics online.

genomics↗

bioRxiv2024

LncRNA-CIR6 Mediates Repair of Infarcted Hearts by Inducing MSCs Differentiation into Cardiogenic Cells through CDK1

PurposeThis study aims to investigate the induction effect of LncRNA-CIR6 on MSCs differentiation into Cardiogenic Cells in vitro and in vivo. MethodsIn addition to pretreatment with the Ro-3306 (CDK1 inhibitor), LncRNA-CIR6 was transfected into BMSCs and hUCMSCs using jetPRIME. LncRNA-CIR6 was transfected into C57BL/6 mice heart by 100 L of AAV9-cTnT-LncRNA-CIR6-ZsGreen i.v. After 3 weeks of transfection followed by AMI surgery, hUCMSCs (5x105/100 L) were injected by i.v 1 week later. Cardiac function was evaluated using VEVO 2100 and electric mapping 9 days after cell injection. IF, Evans blue-TTC, Masson staining, FACS, and WB were used to determine relevant indicators. ResultsLncRNA-CIR6 induced a significant percentage of differentiation in BMSCs (83.00{+/-}0.58)% and hUCMSCs (95.43{+/-}2.13)% into cardiogenic cells, as determined by the expression of cTnT. Compared with MI group, cardiac contraction and conduction function in MI heart treated by LncRNA-CIR6 or combined with MSCs injection groups were significantly increased as well as the area of MI and fibrosis were significantly lower. The transcriptional expression region of LncRNA-CIR6 was in Chr17 from 80209290 to 80209536. The functional region of LncRNA-CIR6 was located at nucleotides 0-50/190-255 in the sequence. CDK1 is a protein found to be related to the proliferation and differentiation of cardiomyocytes is which located in the functional region of LncRNA-CIR6 secondary structure (from 0 to 17). Ro-3306 impeded the differentiation of MSCs into cardiogenic cells, while MSCs transfected with LncRNA-CIR6 showed high expression of CDK1. LncRNA-CIR6 mediates repair of infarcted hearts by inducing MSCs differentiation into cardiogenic cells through CDK1. ConclusionsLncRNA-CIR6 mediates repair of infarcted hearts by inducing MSCs differentiation into cardiogenic cells through CDK1.

molecular biology↗

bioRxiv2024

Identification of a distinct cluster of GDF15high macrophages exhibiting anti-inflammatory activities

Macrophage-mediated inflammatory response may have crucial roles in the pathogenesis of a variety of human diseases. Growth differentiation factor 15 (GDF15) is a cytokine of the transforming growth factor-{beta} superfamily, with potential anti-inflammatory activities. Previous studies observed in human lungs some macrophages which expressed a high level of GDF15. In the present study, we employed multiple techniques, including immunofluorescence, flow cytometry, and single-cell RNA sequencing, in order to further clarify the identity of such GDF15high macrophages. We demonstrated that macrophages derived from human peripheral blood mononuclear cells and rat bone marrow mononuclear cells by in vitro differentiation with granulocyte-macrophage colony stimulating factor contained a minor population ([~]1%) of GDF15high cells. GDF15high macrophages did not exhibit a typical M1 or M2 phenotype, but had a unique molecular signature as revealed by single-cell RNA sequencing. Functionally, GDF15high macrophages were associated with reduced responsiveness to pro-inflammatory activation; furthermore, GDF15high macrophages could inhibit the pro-inflammatory functions of other macrophages via a paracrine mechanism. We further confirmed that GDF15 per se was a key mediator of the anti-inflammatory effects of GDF15high macrophage. Also, we provided direct evidence showing that GDF15high macrophages were also present in other macrophage-residing human tissues in addition to the lungs. Our results suggest that the GDF15high macrophage may represent a distinct cluster of macrophage cells with intrinsic anti-inflammatory functions. The (patho)physiological importance of these cells in vivo warrants further investigation.

immunology↗

bioRxiv2023

Crosstalk between Gut Sensory Ghrelin Signaling and Adipose Tissue Sympathetic Outflow Regulates Metabolic Homeostasis

The stomach-derived orexigenic hormone ghrelin is a key regulator of energy homeostasis and metabolism in humans. The ghrelin receptor, growth hormone secretagogue receptor 1a (GHSR), is widely expressed in the brain and gastrointestinal vagal sensory neurons, and neuronal GHSR knockout results in a profoundly beneficial metabolic profile and protects against diet-induced obesity (DIO) and insulin resistance. Here we show that in addition to the well characterized vagal GHSR, GHSR is robustly expressed in gastrointestinal sensory neurons emanating from spinal dorsal root ganglia. Remarkably, sensory neuron GHSR deletion attenuates DIO through increased energy expenditure and sympathetic outflow to adipose tissue independent of food intake. In addition, neuronal viral tract tracing reveals prominent crosstalk between gut non-vagal sensory afferents and adipose sympathetic outflow. Hence, these findings demonstrate a novel gut sensory ghrelin signaling pathway critical for maintaining energy homeostasis.

neuroscience↗

bioRxiv2023

Histone H1x in mouse ventral hippocampus correlates with, but does not cause behavioral adaptations to stress

Prior research has identified differential protein expression levels of linker histone H1x within the ventral hippocampus (vHipp) of stress-susceptible versus stress-resilient mice. These mice are behaviorally classified based on their divergent responses to chronic social stress. Here, we sought to determine whether elevated vHipp H1x protein levels directly contribute to these diverging behavioral adaptations to stress. First, we demonstrate that stress-susceptible mice uniquely express elevated vHipp H1x protein levels following chronic stress. Given that linker histones coordinate heterochromatin compaction, we hypothesize that elevated levels of H1x in the vHipp may impede pro-resilience transcriptional adaptations and prevent development of the resilient phenotype following social stress. To test this, 8-10-week-old male C57BL/6J mice were randomly assigned to stressed and unstressed groups undergoing 10 days of chronic social defeat stress (CSDS) or single housing respectively. Following CSDS, mice were classified as susceptible versus resilient based on their social interaction behaviors. We synthesized a viral overexpression (OE) vector for H1x and transduced experimental mice with either H1x or control GFP within vHipp. Following viral delivery, we conducted social, anxiety-like, and memory-reliant behavior tests on distinct cohorts of mice. We found no behavioral adaptations following H1x OE compared to GFP controls in susceptible, resilient, or unstressed mice. In sum, although we confirm vHipp protein levels of H1x correlate with susceptibility to social stress, we observe no significant behavioral consequence of H1x OE. Thus, we conclude elevated levels of H1x are correlated with, but are not singularly sufficient to drive development of behavioral adaptations to stress.

neuroscience↗

bioRxiv2023

Intracranial tumors elicit systemic sympathetic hyperactivity that limits immunotherapeutic responses

Intracranial tumors present unique challenges for immunotherapy. These can include both local and systemic modes of immune suppression whose mechanistic underpinnings are incompletely understood. Here, we reveal that tumors harbored intracranially elicit systemic increases to circulating catecholamine levels, with the resultant chronic sympathetic hyperactivity driving T cell dysfunction and limiting immunotherapeutic success. Conversely, treatment with {beta}-adrenergic blockade increases NF-{kappa}B activity in immune cells, restores T cell polyfunctionality, modifies the tumor microenvironment, and licenses immune-based therapies in murine models of glioblastoma (GBM) to extend survival. Extended survival is also observed in GBM patients having received {beta}-adrenergic blockade, as well as in patients with melanoma and lung cancer brain metastases who received {beta}-blockade alongside concomitant immune checkpoint inhibition. While {beta}-blockade also impacts outcomes in the setting of extracranial disease, the benefits are especially pronounced in patients harboring intracranial disease burdens. These data suggest that sympathetic hyperactivity facilitates systemic immune dysfunction in the setting of intracranial tumors, specifically and advance a role for {beta}-adrenergic blockade in licensing immunotherapeutic responses within the intracranial compartment.

immunology↗

bioRxiv2023

MetaGIN: A Lightweight Framework for Molecular Property Prediction

Recent advancements in AI-based synthesis of small molecules have led to the creation of extensive databases, housing billions of small molecules. Given this vast scale, traditional quantum chemistry (QC) methods become inefficient for determining the chemical and physical properties of such an extensive array of molecules. As these properties are key to new drug research and development, deep learning techniques have recently been developed. Here, we present MetaGIN, a lightweight framework designed for molecular property prediction. While traditional GNN models with 1-hop edges (i.e., covalent bonds) are sufficient for abstract graph representation, they are inadequate for capturing 3D features. Our MetaGIN model shows that including 2-hop and 3-hop edges (representing bond and torsion angles, respectively) is crucial to fully comprehend the intricacies of 3D molecules. Moreover, MetaGIN is a streamlined model with fewer than 10 million parameters, making it ideal for fine-tuning on a single GPU. It also adopts the widely acknowledged MetaFormer framework, which has consistently shown high accuracy in many computer vision tasks. Through our experiments, MetaGIN achieves a mean absolute error (MAE) of 0.0851 with just 8.87M parameters on the PCQM4Mv2 dataset. Furthermore, MetaGIN outperforms leading techniques, showcasing superior performance across several datasets in the MoleculeNet benchmark.

bioinformatics↗

bioRxiv2023

MTGCL: Multi-Task Graph Contrastive Learning for Identifying Cancer Driver Genes from Multi-omics Data

Cancer is a complex disease that typically arises from the accumulation of mutations in driver genes. Identification of cancer driver genes is crucial for understanding the molecular mechanisms of cancer, and developing the targeted therapeutic approaches. With the development of high-throughput biological technology, a large amount of genomic data and protein interaction network data have been generated, which provides abundant data resources for identifying cancer driver genes through computational methods. Given the ability of graph neural networks to effectively integrate graph structure topology information and node features information, some graph neural network-based methods have been developed for identifying cancer driver genes. However, these methods suffer from the sparse supervised signals, and also neglect a large amount of unlabeled node information, thereby affecting their ability to identify cancer driver genes. To tackle these issues, in this work we propose a novel Multi-Task Graph Contrastive Learning framework (called MTGCL) to identify cancer driver genes. By using self-supervised graph contrastive learning to fully utilize the unlabeled node information, MTGCL designs an auxiliary task module to enhance the performance of the main task of driver gene identification. MTGCL simultaneously trains the auxiliary task and main task, and shares the graph convolutional encoder weights, so that the main task enhances the discriminative ability of the auxiliary task via supervised learning, whereas the auxiliary task exploits the unlabeled node information to refine the node representation learning of the main task. The experimental results on pan-cancer and some specific cancers demonstrate the effectiveness of MTGCL in identifying the cancer driver genes. In addition, integrating multi-omics features extracted from multiple cancer-related databases can greatly enhance the performance of identifying cancer driver genes, especially, somatic mutation features can effectively improve the performance of identifying specific cancer driver genes. The source code and data are available at https://github.com/NWPU-903PR/MTGCL. Author SummaryIdentifying cancer driver genes that causally contribute to cancer initiation and progression is essential for comprehending the molecular mechanisms of cancer and developing the targeted therapeutic strategies. However, wet-lab experiments are time-consuming and labor-intensive. The advent of high-throughput multi-omics technology provides an opportunity for identifying the cancer driver genes through data-driven computing approaches. Nevertheless, effectively integrating these omics data to identify cancer driver genes poses significant challenges. Existing computational methods exhibit certain limitations. For instance, conventional approaches (e.g., gene mutation frequency-based methods, network-based methods) often focus on a single omics data, while existing deep learning-based methods have not fully utilized the abundant unlabeled node information, so that their identification accuracy is not high enough. Thus, by fully utilizing multidimensional genomics data and molecular interaction networks, we propose a multi-task learning framework (called MTGCL) to identify cancer driver genes. MTGCL synergistically combines graph convolutional neural networks with graph contrastive learning. The experimental results validate the power of MTGCL for identifying cancer driver genes.

bioinformatics↗

bioRxiv2023

SCancerRNA: Expression at the Single Cell Level and Interaction Resource of Non-coding RNA Biomarkers for Cancers

Non-coding RNAs (ncRNAs) participate in multiple biological processes associated with cancer as tumor suppressors or oncogenic drivers. Due to their high stability in plasma, urine, and many other fluids, ncRNAs have the potential to serve as key biomarkers for early diagnosis and screening of cancers. During cancer progression, tumor heterogeneity plays a crucial role, and it is particularly important to understand the gene expression patterns of individual cells. With the development of single-cell RNA sequencing (scRNA-seq) technologies, uncovering gene expression in different cell types for human cancers has become feasible by profiling transcriptomes at the cellular level. However, a well-organized and comprehensive online resource that provides access to the expression of genes corresponding to ncRNA biomarkers in different cell types at the single cell level is not available yet. Therefore, we developed the SCancerRNA database to summarize experimentally supported data on long ncRNA (lncRNA), microRNA (miRNA), piwi-interacting RNA (piRNA), small nucleolar RNA (snoRNA), and circular RNA (circRNA) biomarkers, as well as data on their differential expression at the cellular level. Furthermore, we collected biological functions and clinical applications of biomarkers to facilitate the application of ncRNA biomarkers to cancer diagnosis, as well as monitoring of progression and targeted therapies. SCancerRNA also allows users to explore interaction networks of different types of ncRNAs, and build computational models in the future. SCancerRNA is freely accessible at http://www.scancerrna.com/BioMarker.

bioinformatics↗

bioRxiv2023

Functional characterization of gene regulatory elements and neuropsychiatric disease-associated risk loci in iPSCs and iPSC-derived neurons

Genome-wide association studies (GWAS) have identified thousands of non-coding variants that contribute to psychiatric disease risks, likely by perturbing cis-regulatory elements (CREs). However, our ability to interpret and explore their mechanisms of action is hampered by a lack of annotation of functional CREs (fCREs) in neural cell types. Here, through genome-scale CRISPR screens of 22,000 candidate CREs (cCREs) in human induced pluripotent stem cells (iPSCs) undergoing differentiation to excitatory neurons, we identify 2,847 and 5,540 fCREs essential for iPSC fitness and neuronal differentiation, respectively. These fCREs display dynamic epigenomic features and exhibit increased numbers and genomic spans of chromatin interactions following terminal neuronal differentiation. Furthermore, fCREs essential for neuronal differentiation show significantly greater enrichment of genetic heritability for neurodevelopmental diseases including schizophrenia (SCZ), attention deficit hyperactivity disorder (ADHD), and autism spectrum disorders (ASD) than cCREs. Using high-throughput prime editing screens we experimentally confirm 45 SCZ risk variants that act by affecting the function of fCREs. The extensive and in-depth functional annotation of cCREs in neuronal types therefore provides a crucial resource for interpreting non-coding risk variants of neuropsychiatric disorders.

genetics↗

bioRxiv2023

Aminomutation catalyzed by CO2 self-sufficient cascade amino acid decarboxylases

Molecular editing of an amino group from -position of amino acids to its {beta}-position is of scientific interest and could be economically appealing. Here we reconstructed an in vitro biotransformation pathway composed of two cascade decarboxylases, i.e., aspartate {beta}-decarboxylase and aspartate -decarboxylase, and implemented molecular editing to change -alanine into {beta}-alanine. In it, we discovered a new reaction of aspartate {beta}-decarboxylase that can fix CO2 directly. This cascade enzymatic pathway enabled an aminomutation reaction with 100% carbon atom economy. This work presented the first CO2-fixing biological reaction catalyzed by the amino acid decarboxylases and demonstrated a new means for the molecular editing of -amino acids.

biochemistry↗

bioRxiv2023

Fully automated and integrated proteomics sample preparation platform for high-throughput drug target identification

With the increased demand of large-cohort proteomic analysis, fast and reproducible sample preparation has become the critical issue that needs to be solved. Herein, we developed a fully automated and integrated proteomics sample preparation workflow (autoSISPROT), enabling the simultaneous processing of 96 samples in less than 2.5 hours. Benefiting from its 96-channel all-in-tip operation, protein digestion, peptide desalting, and TMT labeling could be achieved in a fully automated manner. The autoSISPROT demonstrated good sample preparation performances, including >94% of digestion efficiency, nearly 100% of alkylation efficiency, >98% of TMT labeling efficiency, and >0.9 of intra- and inter-batch Pearson correlation coefficients. Furthermore, by combining with cellular thermal shift assay-coupled to mass spectrometry (CETSA-MS), the autoSISPROT was able to process and TMT-label 40 samples automatically and accurately identify the known target of methotrexate. Importantly, taking advantage of the data independent acquisition and isothermal CETSA-MS, the autoSISPROT was well applied for identifying known targets and potential off-targets of 20 kinase inhibitors by automatedly processing 87 samples, affording over a 10-fold improvement in throughput when compared to classical CETSA-MS. Collectively, we developed a fully automated and integrated workflow for high-throughput proteomics sample preparation and drug target identification.

systems biology↗

bioRxiv2023

Expression and molecular regulation of key genes during adipogenesis of neural crest stem cells

Neural crest stem cells (NCSCs) are pluripotent stem cells derived from the "fourth germ layer". Similar to mesenchymal stem cells (MSCs), NCSCs can differentiate into a variety of cell types, such as adipocytes. However, the mechanism of NCSCs adipogenesis remains unclear. Previously, we have revealed that primary cells have neural crest and stem cell properties and can differentiate into adipocytes. Therefore, in this study, the differentially expressed genes (DEGs) of NCSCs at specific time points of adipogenesis were predicted by mRNA sequencing, the key genes of adipogenesis were predicted by principal component analysis, heat map, GO and KEGG enrichment analysis, and the expression of DEGs was verified by RT-qPCR. RNA interference (RNAi) technology was used to inhibit the expression of DEGs, and RT-qPCR and Western blot were used to explore the regulatory mechanism between DEGs. Sequencing results indicated a possible regulatory relationship between C/EBP, C/EBP{beta}, C/EBP{delta} and PPAR{gamma}. The results of RT-qPCR were consistent with those of mRNA sequencing. Combined with RT-qPCR and western blot results, we found that C/EBP{beta} and PPAR{gamma} regulated the transcription of C/EBP during NCSCs adipogenesis, and C/EBP{beta} and PPAR{gamma} formed positive feedback loop.

developmental biology↗

bioRxiv2023

Leading Basic Modes of Spontaneous Activity Drive Individual Functional Connectivity Organization in the Resting Human Brain

Spontaneous activity of the human brain provides a window to explore intrinsic principles of functional organization. However, most studies have focused on interregional functional connectivity. The principles underlying rich repertoires of instantaneous activity remain largely unknown. We apply a novel eigen-microstate analysis to three resting-state functional MRI datasets to identify basic modes that represent fundamental activity patterns that coexist over time. We identify a few (i.e., five) leading basic modes that dominate activity fluctuations. Each of these modes exhibits a distinct functional system-dependent coactivation pattern and corresponds to specific cognitive profiles. In particular, the spatial pattern of the first leading basis mode shows the separation of activity between the default-mode and primary and attention regions. We further reconstruct individual functional connectivity as the weighted contribution of these leading basic modes based on theoretical modelling. Moreover, these leading basic modes capture sleep deprivation-induced changes in brain activity and interregional connectivity, primarily involving the default-mode and task-positive regions. Our findings reveal a dominant set of basic modes of spontaneous activity that reflect multiplexed interregional coordination and drive conventional functional connectivity, furthering the understanding of the functional significance of spontaneous brain activity.

neuroscience↗

bioRxiv2023

Co-engagement of TIGIT+ immune cells to PD-L1+ tumours by a bispecific antibody potentiates T-cell response and tumour control

Co-targeting PD-1/PD-L1 and TIGIT/CD226 is being pursued to broaden the efficacy of current immunotherapy. Here we demonstrate that a bispecific antibody (BsAb) targeting PD-L1 and TIGIT, HB0036, shows major advantages over the combination of the two parental monoclonal Abs (mAbs). We demonstrated that HB0036 co-engages PD-L1+ tumour cells and TIGIT+ T cells, and thereby upregulates CD226 on T cells and induces a greater T-cell proliferative response compared to the combination of the parental antibodies in vitro. In vivo, HB0036 recruits greater amounts of TIGIT antibody in PD-L1+ tumours but not in PD-L1- tumours, compared with therapy using the two parental antibodies. We also observed improved tumour control and favourable immunological signatures with HB0036 in syngeneic and xenograft tumour models. Collectively, these findings demonstrate that bispecific antibodies targeting PD-L1 and TIGIT offer superior benefits in cancer immunotherapy compared with therapy using the two parental antibodies. Based on these studies, a phase I clinical trial with HB0036 has been initiated in patients with solid tumours (NCT05417321).

immunology↗

bioRxiv2023

Neural representation strength of predicted category features biases decision behavior

Theories of prediction-for-perception propose that the brain predicts the information contents of upcoming stimuli to facilitate their perceptual categorization. A mechanistic understanding should therefore address where, when, and how the brain predicts the stimulus features that change behavior. However, typical approaches do not address these predicted stimulus features. Instead, multivariate classifiers are trained to contrast the bottom-up patterns of neural activity between two stimulus categories. These classifiers then quantify top-down predictions as reactivations of the category contrast. However, a category-contrast cannot quantify the features reactivated for each category-which might be from either category, or both. To study the predicted category-features, we randomly sampled features of stimuli that afford two categorical perceptions and trained multivariate classifiers to discriminate the features specific to each. In a cueing design, we show where, when and how trial-by-trial category-feature reactivation strength directly biases decision behavior, transforming our conceptual and mechanistic understanding of prediction-for-perception.

neuroscience↗